Fault Detection Diagnostics: a critical element in building analytics suite of applications

Remember the incident when Tony Stark and Rhodey were at the restaurant discussing the secrecy around Mandarin and his recent mischiefs? Suddenly Tony starts feeling uneasy and steps out to get into his Iron man suit. He then asks Jarvis to check his heart and brain activities. Jarvis does a quick check on his vital signs and suggests that he is experiencing an anxiety attack. 

Now note that he never had an anxiety attack before. Hence, there were no historical conditions that existed to classify it as a symptom of anxiety yet Jarvis detected it accurately. This is perhaps the closest real-life example of what Fault detection diagnostics is and it’s marvelous benefits.

One may go deeper to point out that Jarvis might have collected a similar vital data pattern from the internet. Then came up with the logic to classify it as an anxiety attack, an excellent case of learning a new fault scenario. Well, let’s leave it for geeks to interpret it further.   

In the context of commercial buildings analytic solutions, Fault detection diagnostics (FDD) identifies anomalies in the operation of critical assets such as HVAC networks, elevators, or boilers and notify the operator along with the root cause analysis. If we break down the term into two parts then fault detection can be considered as a process of pinpointing anomalous behavior in the asset while the diagnosis is a process of isolating probable causes of the fault after getting detected. 

For example, in Air Handling Unit (AHU) sensible wheel gradually started running at 60% of its rated speed, which was higher than the normal condition. An FDD noticed the change in behavior and started examining other parameters to come up with the diagnosis that the set point for the cooling coil was fixed at lower than the optimum value. This was causing temperature disparity and excess energy loss in heating up the already cooled air to the desired Supply Air temperature. 

Why Fault detection diagnostics (FDD) stands out?

Blind spots:

Now what differentiates an FDD from other basic logic in Building management systems (BMS) or Building automation systems (BAS) is its ability to identify the ‘blind spots’ in the operational behavior of an asset. In the above example, BMS couldn’t flag the issue since the operator had fixed the setpoint to a lower temperature. In fact, BMS is not supposed to give any insights as fundamentally it is not built for that. Whereas FDD can examine such human interventions and suggest corrective action. 

Contextualization:  

It is not just about creating a logic that alerts an operator once the threshold value is crossed. FDD can collect contextual information about the entire network of influencing asset parameters and then judge the change in the asset behavior. If the blowers in AHU start drawing more current than the rated value that could be a sign of an electrical fault or a clogged air filter. One can tell the exact reason only after contextualizing the issue with the surrounding conditions. 

Library of fault scenarios:

A highly experienced technician can remember the probable causes of the faults that occur in an elevator. But his knowledge or understanding is limited to him only, making his facility management team dependent on him to monitor and manage that particular asset. With the changing occupancy rate and adapting to a new tech stack assets can come under entirely new operational scenarios. Something even an experienced technician has never perceived. Fortunately, an FDD algorithm can automatically run multiple scenarios and suggest the team if anything is about to go wrong.        

There are different mathematical models, methods and approaches available to classify faults and detection analysis. But usually, three distinct approaches find their way into an ideal FDD application: Quantitative model, qualitative model and process history-based. However recently, a hybrid approach is gaining momentum as it eliminates the loopholes in individual approaches.  

Berkeley Lab published a report on FDD, in this report it suggested that using FDD tools and adopting efficiency measures based on FDD findings can save 5-30% of energy. As a facility manager, once you understand the relevance and benefits of FDD for your building management. The next and probably the most decisive question arises, how to select an ideal FDD application for your building? Should you go for the product or service. 

We will cover that discussion in the next article. Till then let us know your journey to explore or if you are already using it then share your experience with us. What kind of building analytics solution are you using? How does it help to optimize your building management? Share with us!